Optimization method for improving imaging quality of single-pixel camera through image reconstruction algorithm based on deep learning

Abstract

The invention discloses an optimization method for improving the imaging quality of a single-pixel camera through an image reconstruction algorithm based on deep learning. With the development of a deep convolutional neural network (DCNN), the application of the DCNN in the field of super-resolution imaging also develops. Based on the basic principles of the compressed sensing and the single-pixel camera imaging, a deep learning network structure model for image super-resolution reconstruction is designed, a novel deep learning image reconstruction algorithm is embedded into a single-pixel imaging system, and a deep learning technology and a single-pixel camera super-resolution imaging technology are combined. Compared with a traditional matching pursuit algorithm, a minimum L1 norm algorithm and an iteration threshold algorithm for compressed sensing image reconstruction, the novel deep learning algorithm effectively improves the image reconstruction precision and the imaging quality and effect of the single-pixel camera. The effectiveness of the imaging optimization of the single-pixel camera is verified through the simulation and the actual imaging experiments in the deep learning manner.

Description

technical field [0001] The invention belongs to the field of signal restoration and reconstruction and intelligent computing in signal and information processing, and particularly relates to an optimization method for improving the imaging quality of a single-pixel camera based on a deep learning-based image reconstruction algorithm. Background technique [0002] Donoho, Candès and others put forward the theory of compressed sensing in 2006. The core idea of ​​compressed sensing theory is to combine sampling and compression in signal sampling and compression theory. The main content of compressed sensing is based on the premise that the original signal is sparse or can be represented sparsely. The linear projection value of the original signal is obtained through the measurement matrix with a sampling frequency lower than the Nyquist sampling theorem, and the compressed representation of the signal is directly obtained. The reconstruction algorithm restores and reconstructs ...

AI Technical Summary

Problems solved by technology

[0004] The technical problem to be solved by the present invention is: in traditional compressed sensing signal recovery and single-pixel camera imaging, image reconstruction and imaging often require an efficient signal recovery algorithm to reconstruct images, but traditional image reconstruction algorithms often have reconstruction problems It takes a long time and low efficiency, it is difficult to find the global optimal solution, and the quality and accuracy of the reconstructed image cannot meet the requirements

Images